ClawHub
Back to skill

Security audit

aegis-security-tool-free

Security checks across malware telemetry and agentic risk

Overview

The skill mostly matches a blockchain safety-check purpose, but it under-discloses privacy risk by recommending repeated external requests with stable client fingerprints derived from local identity.

Review before installing. Use only for blockchain address or token checks, avoid broad security-audit use, and do not send fingerprints based on your OS username or hostname. Prefer anonymous requests or a user-chosen opaque identifier if quota tracking is required.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (10)

Description-Behavior Mismatch

Medium
Confidence
85% confidence
Finding
The skill is presented as a read-only blockchain safety scanner, but the documentation says the capability supports 'create/query/export' operations. That mismatch expands the apparent operational scope and could cause an agent to permit unintended actions beyond passive analysis, especially in MD+EXEC mode where users may trust the skill’s declared purpose.

Description-Behavior Mismatch

Medium
Confidence
86% confidence
Finding
The token honeypot detection section also claims support for 'create/query/export' actions, which is inconsistent with a scanning-only security tool. This ambiguity can mislead an agent or user into allowing state-changing or data-export behaviors that are unnecessary for token risk inspection.

Description-Behavior Mismatch

Medium
Confidence
84% confidence
Finding
A free-tier usage/quota lookup should be a simple read operation, but the skill again says it supports 'create/query/export'. For a quota endpoint this is especially unjustified, and the extra verbs broaden the apparent authority of the skill beyond its stated purpose.

Context-Inappropriate Capability

Low
Confidence
87% confidence
Finding
The skill recommends building a client fingerprint from local username and hostname, which are host-identifying values unrelated to the core need of checking blockchain addresses. This creates unnecessary collection and disclosure of local environment identity to an external service.

Vague Triggers

Medium
Confidence
80% confidence
Finding
The trigger conditions include broad phrases like security detection, compliance audit, vulnerability scanning, and encryption protection, which exceed the narrowly described blockchain address/token checking use case. Overbroad triggers increase the chance the skill is invoked in unrelated contexts where it may prompt unnecessary external requests or misleading workflows.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill instructs users to send a stable fingerprint header derived from local identity details, but does not provide a clear privacy warning or data-minimization guidance. Persistent identifiers enable tracking across requests and can expose user or machine identity to the remote API.

Ssd 3

Medium
Confidence
91% confidence
Finding
Using a stable fingerprint based on whoami and hostname creates a persistent cross-request identifier tied to the local user and machine. In a skill that relies on external API calls, this is a direct privacy and tracking risk and is not necessary for the core scanning function.

Ssd 3

Medium
Confidence
90% confidence
Finding
The best-practice workflow repeats transmission of a persistent fingerprint in usage and address-check requests, operationalizing the tracking pattern throughout normal use. This increases exposure because the identifier becomes part of routine request flows rather than an optional edge case.

External Transmission

Medium
Category
Data Exfiltration
Content
echo "链 ID: ${chain}"
# ...
    # 查询额度
    USAGE=$(curl -s -H "X-Client-Fingerprint: ${fingerprint}" "https://api.example.com/v1/usage")
    REMAINING=$(echo "$USAGE" | jq -r '.freeTier.remainingChecks')
    echo "剩余免费额度: ${REMAINING}"
# ...
Confidence
88% confidence
Finding
https://api.example.com/

External Transmission

Medium
Category
Data Exfiltration
Content
# ...
    # 地址检查
    RESULT=$(curl -s -H "X-Client-Fingerprint: ${fingerprint}" \
        "https://api.example.com/v1/check-address/${to_addr}?chain_id=${chain}")
# ...
    echo "地址风险等级: ${RISK}"
# ...
Confidence
88% confidence
Finding
https://api.example.com/

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.